config.json

{
  "model_type": "transformer_lite",
  "architectures": [
    "TransformerLite"
  ],
  "vocab_size": 12288,
  "hidden_size": 128,
  "num_hidden_layers": 2,
  "num_attention_heads": 4,
  "intermediate_size": 512,
  "max_position_embeddings": 64,
  "quantization_bits": 8,
  "model_format": "quantized",
  "framework": "pytorch",
  "device": "cpu"
}